Talky Labs

Cover Block

From the public record

The narrow point of orientation here is straightforward: Talky Labs is a newly funded Madrid startup trying to automate accounting and operational finance with AI, and nearly every public fact available today traces back to the company and one financing report [Expansion, October 2026].

Field	Value
Name	Talky Labs
Tagline	AI software for automating accounting and operational finance workflows for businesses.
Headquarters	Madrid, Spain [Expansion, October 2026]
Founded	2025 [Expansion, October 2026]
Stage	Pre-Seed [Expansion, October 2026]
Business Model	SaaS
Industry	Fintech
Technology	AI / Machine Learning
Geography	Western Europe
Growth Profile	Venture Scale
Founding Team	Jorge Beamonte and Maximilian Letón [Expansion, October 2026]
Funding Label	Pre-seed [Expansion, October 2026]
Total Disclosed	€1.5 million [Expansion, October 2026]

Links

From the public record

• Website: https://usetalky.com/careers

The Short Version

PUBLIC Talky Labs is building AI software for accounting and operational finance, and it merits attention now because it has moved quickly from a 2025 founding to a reported €1.5 million pre-seed round and early workflow volume claims in a category where real automation, rather than simple data capture, remains difficult to prove [Expansion, October 2026] [Crunchbase, retrieved 2026]. Founded in Madrid by Jorge Beamonte and Maximilian Letón, the company describes its system as an automation layer that learns a customer’s historical accounting treatment and finance team judgment, then uses that context to centralize knowledge, define workflows, automate finance operations, and resolve exceptions with traceability and auditability [Expansion, October 2026] [Talky, October 2026].

The product pitch is notable less for generic AI positioning than for the claim that it can process between 86% and 93% of documents without human intervention, while also capturing invoices and business documents, proposing accounting entries, reconciling bank movements, and preparing structured outputs for existing ERP systems, according to company materials and Expansion’s reporting [Expansion, October 2026] [Talky, October 2026]. Those performance and scale figures, including more than 1.8 million processed documents representing more than €2 billion in customer spending by October 2026, are company-reported and should be treated cautiously until independently corroborated [Expansion, October 2026].

On the team side, Beamonte is identified as co-founder and CEO, and Letón as co-founder and operations director, with several hires visible on LinkedIn in AI financial engineering and business development roles during 2026, which suggests the company is already staffing both product and go-to-market functions at a very early stage [Expansion, October 2026] [LinkedIn, retrieved 2026]. The financing appears to be a pre-seed round backed by HelloWorld, Silence VC, Motive Force, and named angels including Sebastián Gunningham, Paul Golbaum, and Javier Morenés, with the business model described as SaaS [Expansion, October 2026] [Crunchbase, retrieved 2026].

Over the next 12 to 18 months, the main points to watch are whether Talky can convert its document-processing narrative into visible customer proof, maintain automation quality across varied accounting edge cases, and turn an apparently implementation-heavy product motion into repeatable software economics; the current hiring for a Forward Deployed Engineer and AI Core Engineer points to both opportunity and execution complexity [Talky, October 2026]. The setup is promising, but most of the operating evidence in public view still comes from the company or a single reported financing event rather than broad third-party validation [Expansion, October 2026] [Talky, October 2026].

Unconfirmed -- This section relies materially on company claims and a single primary news report, with partial corroboration from Crunchbase and LinkedIn.

Taxonomy Snapshot

Axis	Value
Stage	Pre-Seed
Business Model	SaaS
Industry / Vertical	Fintech
Technology Type	AI / Machine Learning
Geography	Western Europe
Growth Profile	Venture Scale
Founding Team	Co-Founders (2)
Funding	Pre-seed

The Company in Brief

PUBLIC

Talky Labs enters the public record as a very young Madrid fintech, and the verifiable facts are still sparse. Crunchbase lists the company as Talky Labs, headquartered in Madrid, Spain, founded in 2025, and categorized in software and finance-oriented startup databases as a private company at pre-seed stage [Crunchbase]. The company website describes Talky as building AI software for accounting teams and firms, with a product aimed at automating accounting and operational finance workflows while preserving control, traceability, and auditability [Talky | AI for accounting teams and firms, retrieved 2026].

The founding record is straightforward, if thin. Public reporting identifies Jorge Beamonte as co-founder and CEO and Maximilian Letón as co-founder and operations director, with Beamonte stating the company was founded in 2025 [Crunchbase] [Talky | AI for accounting teams and firms, retrieved 2026]. The public milestone sequence is short: formation in 2025, early team build-out visible through 2026 hiring activity on the careers page, and a pre-seed financing disclosed in October 2026 [Crunchbase] [Talky Careers, October 2026]. That timeline suggests a company still in its first institutional chapter rather than one with a long operating history.

Single-source, plausible -- Confirmed in part by Crunchbase and the company website; some founding-team role detail remains primarily company or database sourced.

What They Have Built

Mixed sourcing

Talky is pitching a narrower, and potentially more defensible, automation layer than generic document capture. Public descriptions from Expansion and the company site converge on the same core claim: the software is built for accounting and operational finance workflows, and it is meant to learn a company’s historical accounting treatment along with the judgment calls its finance team makes in edge cases [Expansion, October 2026] [Talky, October 2026]. In practice, the platform is described as centralizing finance knowledge, defining workflows, automating finance operations, and resolving exceptions while preserving control and traceability, which suggests the product is being positioned as a system of decision support and workflow execution rather than a standalone OCR tool [Expansion, October 2026] [Talky, October 2026].

The workflow surface that is actually described in public is specific enough to matter. According to Talky’s website, the platform captures invoices and other business documents, proposes accounting entries, reconciles bank movements, and prepares structured outputs for existing ERP systems [Talky, October 2026]. Expansion adds a quantitative company claim that the software can process between 86% and 93% of documents without human intervention, while also reporting more than 1.8 million documents processed and more than €2 billion in customer spending represented by those documents as of October 2026 [Expansion, October 2026]. Those operating figures are useful as directional evidence of product ambition, but they remain company-reported and should be read as unverified external claims rather than independently audited performance data [Expansion, October 2026].

The technical stack is mostly opaque in public materials, so the safest read comes from role design rather than architecture claims. Talky’s October 2026 careers page shows hiring for an AI Core Engineer and a Forward Deployed Engineer, with the latter based in Madrid and customer sites, which supports an interpretation that implementation may involve close customer-specific workflow tuning and on-site deployment work (inferred from job postings) [Talky, October 2026]. That fits the product narrative around encoding firm-specific accounting judgment, although no verified demo, model architecture, or ERP integration list is publicly described in the available sources [Talky, October 2026] [Expansion, October 2026].

Unconfirmed -- Material product capabilities and performance figures rely primarily on company descriptions and one press report, with no independent technical verification [Expansion, October 2026] [Talky, October 2026].

Market Size and Demand

PUBLIC

The market matters now because finance teams are under pressure to automate higher-volume back-office work, but the public record for Talky points to a narrower wedge than broad "finance AI": company-specific accounting judgment encoded into workflow software [Expansion, October 2026].

Public sources do not provide a direct TAM, SAM, or SOM for Talky, and there is no cited third-party market study in the supplied record that cleanly sizes AI accounting automation in Spain or Western Europe. The available reporting instead supports a category definition: software for accounting and operational finance automation that sits between source documents, internal review steps, and existing ERP systems [Expansion, October 2026] [Talky, October 2026]. That framing matters because it places Talky closer to workflow automation for controllership functions than to consumer fintech or generic document OCR, with demand likely tied to the volume and variability of accounting operations rather than a single vertical [Expansion, October 2026] [Talky, October 2026].

Demand signals in the public record are mostly indirect, but they are consistent. Expansion reported that Talky positions its product around learning a company's historical accounting treatment and finance team's judgment, then using that to automate finance operations and exception handling with traceability [Expansion, October 2026]. The company website describes invoice and business-document capture, suggested accounting entries, bank-movement reconciliation, and structured outputs for existing ERPs [Talky, October 2026]. Read conservatively, that suggests buyer interest where manual review remains expensive, audit trails matter, and finance teams cannot easily standardize every edge case across entities or workflows [Expansion, October 2026] [Talky, October 2026].

The adjacent markets are easier to identify than the core market's size. One substitute path is point automation for document extraction, invoice capture, or bank reconciliation, where the value proposition is labor reduction at a single step rather than system-level workflow control [Talky, October 2026]. Another is broader ERP modernization, where finance automation is bundled into larger software rollouts rather than purchased as a standalone layer [Talky, October 2026]. Talky's own language suggests it is trying to occupy the middle ground: an automation layer that plugs into existing finance systems while preserving controls and auditability, which can be attractive for teams that are not ready to replace core ERP infrastructure [Talky, October 2026].

Macro and regulatory forces also fit the category, even if the supplied sources do not quantify them. Accounting workflows are unusually sensitive to error rates, reproducibility, and evidence trails, so automation products in this segment are likely to be judged less on headline model capability than on whether every decision can be traced back to a source document and reviewed when exceptions arise [Expansion, October 2026] [Talky, October 2026]. That is one reason the reported claim of processing 86% to 93% of documents without human intervention should be read as directional rather than market-defining until independently corroborated, because adoption in finance usually depends on controls, integration effort, and risk tolerance as much as raw automation rates [Expansion, October 2026].

Market lens	What the sources support	Source
Core category	AI software for accounting and operational finance workflows	[Expansion, October 2026]
Product boundary	Captures invoices and business documents, proposes accounting entries, reconciles bank movements, and prepares outputs for ERPs	[Talky, October 2026]
Buyer problem	Reduce manual review while keeping control, traceability, and auditability	[Talky, October 2026]
Market wedge	Learns historical accounting treatment and finance-team judgment to automate exceptions and workflows	[Expansion, October 2026]

The table underscores a familiar early-stage pattern: the category is legible, but market sizing is still inferential from workflow pain points rather than anchored in a named third-party report. For investors, that shifts the burden from TAM storytelling to proof that the product can sustain automation in real production finance environments [Expansion, October 2026] [Talky, October 2026].

Unconfirmed -- This section relies mainly on company descriptions and a single reported article, with no independent third-party market sizing study in the supplied public sources.

Who Else Is Fighting for This

MIXED Talky appears to be positioning itself less as a general bookkeeping tool and more as an automation layer for finance teams and accounting firms that want company-specific accounting judgment encoded into workflow, a narrower claim than broad ERP replacement and a more ambitious one than basic document capture [Expansion, October 2026] [Talky, October 2026].

The public record is thin on named direct rivals, which matters because competitive position can only be assessed by category shape rather than verified head-to-head comparisons. On the incumbent side, the practical alternatives are existing ERP and accounting stacks that already sit inside the finance workflow, since Talky says its software prepares structured outputs for existing ERPs rather than replacing them [Talky, October 2026]. On the challenger side, the clearest overlap is with accounting automation vendors that begin with invoice capture, bank reconciliation, and workflow orchestration, because Talky publicly claims all three surfaces in some form, from capturing invoices and business documents to proposing entries and reconciling bank movements [Talky, October 2026]. Adjacent substitutes also matter here: accounting firms can still solve the same problem with offshore labor, internal process design, or point solutions stitched together around the ERP, and that substitute set is often the first benchmark a buyer uses before adopting a new AI layer.

What looks differentiated in the available reporting is the company-specific learning claim. Expansion reports that the platform learns a customers historical accounting treatment and the finance teams judgment, while the company site emphasizes control, traceability, auditability, and linkage back to source documents [Expansion, October 2026] [Talky, October 2026]. If that is true in production, the edge is not the model layer by itself but the workflow memory built from prior accounting decisions. That can compound over time as more edge cases are resolved. It is also perishable. A better-capitalized incumbent or adjacent automation vendor could make a similar product argument, and the public evidence does not yet show proprietary distribution, exclusive data access, regulatory approvals, or customer lock-in that would make the position hard to copy.

The exposure is easier to name than the moat. Because Talky integrates into existing ERP environments, it likely depends on systems of record and does not own the primary financial workflow, at least based on public product descriptions [Talky, October 2026]. That leaves it exposed to two pressures. First, incumbent accounting platforms can move downward into automation features while keeping distribution inside the core ledger. Second, specialist AI accounting tools with deeper product breadth could compete on implementation confidence if buyers prefer vendors with a longer operating history. Public reporting also does not identify named customers, channel partners, or major accounting firm relationships, so there is no verified evidence yet that Talky controls a repeatable go-to-market wedge beyond the product narrative [Expansion, October 2026].

The most plausible 18-month scenario is a category sort between embedded incumbents and workflow-layer specialists. If finance buyers continue to prefer modular tools that sit on top of existing ERPs, Talky is a potential winner if its reported automation rate of 86 percent to 93 percent holds under broader deployment and remains auditable in practice [Expansion, October 2026]. If buyers instead consolidate around incumbent finance software with native AI features, the loser is likely any stand-alone workflow layer without owned distribution. On the evidence available, Talky fits that risk profile more than the incumbent side. That does not weaken the product claim, but it does mean competitive durability will probably be decided by implementation speed, reference accounts, and depth inside existing finance processes rather than by model performance alone.

Opportunity

PUBLIC

The prize here is unusually large for a company this young because accounting remains one of the most repetitive, judgment-heavy functions inside every business, and Talky is positioning itself not as another extraction tool but as the software layer that learns how a finance team actually makes decisions [Expansion, October 2026] [Talky, October 2026].

The headline opportunity is to become the operating system for operational finance inside mid-market and enterprise teams, first in Spain and then more broadly across Europe. That is a bigger outcome than invoice capture or bookkeeping assistance, because the company says its platform learns a customer's historical accounting treatment, centralizes that knowledge, proposes entries, reconciles bank movements, and prepares structured outputs for existing ERPs, all while keeping control, traceability, and auditability in place [Expansion, October 2026] [Talky, October 2026]. If those claims hold in production, the reachable outcome is not simply labor reduction. It is control of a high-frequency workflow where every processed document improves the system's understanding of customer-specific accounting logic, which tends to sit close to the ledger and close to the ERP, two places where software can become sticky quickly [Expansion, October 2026] [Talky, October 2026].

The current public evidence is still thin and heavily company-sourced, but it does show early signs of scale in the underlying workflow. Expansion reported that by October 2026 Talky had processed more than 1.8 million documents tied to more than €2 billion in customer spending, and the company said it could process between 86 percent and 93 percent of documents without human intervention [Expansion, October 2026]. Those figures are not independently corroborated, so they should be handled conservatively. Still, if directionally accurate, they suggest the product is already being exposed to meaningful document volume, which is the raw material needed to train workflow-specific automation rather than a generic finance assistant [Expansion, October 2026].

Scenario	What happens	Catalyst	Why it’s plausible
ERP-adjacent automation layer	Talky becomes the preferred automation layer between source documents and incumbent finance systems for Spanish and then European accounting teams.	A successful set of ERP integrations and repeatable deployment motion for finance teams that need structured outputs rather than full rip-and-replace software [Talky, October 2026].	The product is already described as preparing structured outputs for existing ERPs, which lowers the adoption hurdle versus replacing the system of record [Talky, October 2026].
Embedded standard for accounting firms	Talky wins accounting firms as a distribution channel, with firms using it across many client workflows and carrying the product into adjacent markets.	A category-tipping reference customer or partner in the accounting-firm channel, where one deployment can expose the product to many downstream entities [Talky, October 2026].	The company explicitly says it serves both accounting firms and finance teams, which gives it a plausible multi-tenant distribution wedge beyond direct enterprise sales [Talky, October 2026].
Judgment engine for operational finance	Talky moves beyond document handling into the default decision layer for repetitive accounting judgments and exception resolution.	Proving that customer-specific historical treatment can be captured and reused with auditable controls at high automation rates [Expansion, October 2026].	Talky's public positioning rests on learning the unwritten rules of accounting and resolving exceptions with traceability, which is a broader claim than OCR or AP automation alone [Expansion, October 2026] [EU-Startups, October 2026].

What compounding could look like is fairly clear, even if the proof is early. Each processed document gives the system another example of how a given company classifies transactions, reconciles movements, and handles edge cases; each accepted suggestion should reduce future manual review; and each workflow connected to an existing ERP increases switching costs because the software becomes part of the customer's control environment [Expansion, October 2026] [Talky, October 2026]. That is not a classic consumer network effect. It is a workflow flywheel: more volume produces better suggestions, better suggestions improve automation rates, stronger automation justifies deeper product adoption, and deeper adoption yields more embedded knowledge of the finance stack. The careers page adds a useful clue here. Hiring for a Forward Deployed Engineer and an AI Core Engineer suggests the company is still pairing model development with hands-on implementation, a pattern often seen when a startup is trying to turn bespoke wins into a repeatable deployment engine [Talky, October 2026].

The size of the win is best framed conditionally because no public market-sizing data or direct public comparable is supplied in the source set. On the evidence available, the upper-bound scenario is that Talky becomes a European control layer for accounting automation, sitting upstream of ERPs and downstream of source documents across finance teams and accounting firms [Expansion, October 2026] [Talky, October 2026]. If that scenario plays out, the company would resemble a system-of-work platform rather than a narrow point solution, which is the kind of software category that can support venture-scale outcomes. A concrete valuation range would be speculative without verified revenue, retention, or category comparable data in the public record. The more defensible statement is that the architecture of the opportunity is large enough to support an outlier result if the reported automation performance proves durable in production and if the company can convert early workflow intensity into repeatable distribution [Expansion, October 2026] [EU-Startups, October 2026].

Unconfirmed -- This section relies materially on company claims reported by Expansion and described on Talky's own website, with limited independent public corroboration beyond financing coverage.

Sources

From the public record

1. [Expansion, October 2026] Talky cierra 1,5 millones de euros para agilizar la contabilidad de las compañias | https://www.expansion.com/expansion-empleo/emprendedores/2026/10/29/talky-cierra-1-5-millones-euros-agilizar-contabilidad-companias.html

2. [Crunchbase, retrieved 2026] Crunchbase | Private Company Data & Predictive Intelligence | https://crunchbase.com/funding_round/talky-labs-pre-seed--cbd09559

3. [Talky, October 2026] Talky Careers | https://usetalky.com/careers

4. [LinkedIn, retrieved 2026] Mert Demirel - EDEN Fund | LinkedIn | https://www.linkedin.com/in/mert-demirel-/

5. [EU-Startups, October 2026] Madrid's Talky raises €1.5 million to teach AI the unwritten rules of accounting | https://www.eu-startups.com/2026/10/madrids-talky-raises-e1-5-million-to-teach-ai-the-unwritten-rules-of-accounting