The enterprise automation stack is full of dead ends. An AI agent can be programmed to book a flight on a modern consumer website, but ask it to pull a patient record from a twenty-year-old hospital portal or submit an invoice through a legacy procurement system, and it hits a wall. The problem isn't the agent's intelligence, it's the interface,or lack of one. Pomerado, a two-person startup out of Y Combinator's F26 batch, is building a bridge over that wall. Its product turns any website, legacy portal, or internal tool into a structured API or Model Context Protocol (MCP) server in minutes, aiming to wire AI agents directly into the brittle, API-less systems that run much of the corporate world [Pomerado].
For founders Akshay Manglik and Alan Ma, the wedge is clear. Their prior work driving enterprise agent strategy and deployments at Fortune 500 companies across finance, healthcare, and software showed them the bottleneck firsthand [Pomerado]. They cite specific experience with deployments involving the Mayo Clinic, where legacy electronic health record (EHR) systems presented a formidable integration challenge [Y Combinator]. Pomerado's pitch is that it can convert those legacy systems into what it calls "AI-native components automatically," bypassing the need for costly, time-consuming custom API development [Y Combinator].
The Wedge Into the Back Office
The target is the sprawling back-office landscape of ERP, EHR, and CRM systems, many of which were built before the API era. Pomerado's documentation illustrates the concept with tools for everyday web tasks, like searching Google Flights or comparing prices on Instacart [Pomerado, October 2026]. The translation to enterprise use is straightforward: an agent could be tasked with checking inventory levels in a legacy SAP module, updating a patient record in an old Cerner system, or pulling a sales report from a custom-built internal dashboard. The company argues this enables faster customer integrations, enterprise assistants connected to internal tools, and more capable personal assistants [Pomerado].
The core of the claim is performance. In a competitive landscape that includes browser-based agent tools like Astra Computer Use, Instinct, and Muse, Pomerado states its approach is 2x faster, 10 to 1,000 times cheaper, and 30% more accurate for browser use [Pomerado]. An independent post on X from investor Aaron Epstein echoed the claim, stating Pomerado builds APIs that are faster (2-4x), cheaper (10-1000x), and more reliable (+30%) than browser agents [Aaron Epstein on X]. The company's core integration minting harness is open source and MIT licensed, a move likely intended to drive developer adoption and scrutiny [GitHub].
The Team Behind the Deployment Experience
The founders bring a specific blend of consulting and applied AI engineering that fits the enterprise sales motion they'll need. The team is composed of former McKinsey and Scale AI employees [Pomerado].
| Role | Name | Prior Experience |
|---|---|---|
| Co-Founder & CEO | Akshay Manglik | Applied AI Engineer, Scale AI; AI Intern, PayPal; ML Research, Johns Hopkins APL [Company Launch Tracker, September 2026]. |
| Co-Founder & CTO | Alan Ma | Former McKinsey [Pomerado]. |
Manglik's background at Scale AI, a company that built a business on data labeling for AI, and Ma's experience at McKinsey, where he would have seen countless legacy system challenges, form a pragmatic foundation. Their stated experience is in strategy and deployment, not just product building, which suggests an understanding of the procurement and integration cycles they'll face [Pomerado].
The Open-Source Play and Early Traction
Pomerado's decision to open-source its core technology is a calculated go-to-market bet. It lowers the barrier to trial for developers and technical teams who can clone the repository and test it against their own internal tools. The company publicly uses Porter for its own scalable, SOC 2-compliant infrastructure, indicating a focus on the enterprise-grade security and reliability that will be non-negotiable for its target customers [LinkedIn]. While the company is early,it has two employees and is part of the current YC batch,its traction is framed through prior deployment experience rather than a long list of current paying customers [Y Combinator]. The referenced work with Fortune 500 companies and the Mayo Clinic speaks to the founders' credibility and the problem's urgency, but it does not yet establish a recurring revenue stream from those organizations [Y Combinator, Pomerado].
Where the Integration Challenge Gets Real
The ambition is significant, but the path is lined with the gritty realities of enterprise IT. Pomerado's performance claims are compelling, but they are company-reported benchmarks against a set of competitors that are also early-stage [Pomerado]. The real test will be in production environments with unpredictable legacy systems, where reliability and accuracy are measured in business continuity, not lab speed.
The competitive set is also evolving rapidly. While Pomerado positions itself against browser-automation agents, its deeper competition may come from other angles:
- Legacy system modernization platforms. Companies like Workato or UiPath that offer broader integration and RPA capabilities could extend their own agent features.
- Major cloud providers. AWS, Google, and Microsoft could eventually bake similar website-to-API capabilities into their agent toolkits as a feature, not a product.
- Internal development. Large enterprises with deep engineering benches may still choose to build custom integrations for their most critical systems, viewing them as core IP.
Pomerado's answer likely hinges on its focus and its open-source wedge. By being singularly focused on this translation layer and giving away the core tools, it hopes to become the default standard for agent-to-website communication before larger platforms can react.
The Next Twelve Months
For a company at this stage, the immediate milestones are straightforward but critical. Securing its first handful of design-partner customers who will implement Pomerado in a live, complex environment will be the most important validation. The founders' prior deployment experience should shorten that sales cycle. The next likely step is a priced seed round to build out a commercial team around the open-source project and develop the enterprise features,audit logs, SSO, advanced monitoring,that turn a developer tool into a department-level purchase.
The ideal customer profile is a mid-to-large enterprise with a mix of modern and legacy systems, a growing investment in AI agent pilots, and a technical operations or business automation team frustrated by integration bottlenecks. Think a healthcare provider with old EHRs, a financial services firm with proprietary trading portals, or a manufacturer with custom supply chain dashboards.
The realistic competitive set for a procurement officer evaluating this category today isn't just other AI startups. It's the build-versus-buy calculation for a custom API, the cost of a broader RPA platform, or the decision to wait for a feature from a cloud giant. Pomerado's bet is that its speed, cost, and accuracy advantages, coupled with the adoption velocity of its open-source core, will make that calculation tilt in its favor. The next year will be about proving that math works outside the lab, inside a real company's back office.
Sources
- [Pomerado] Pomerado, https://pomerado.ai/
- [Y Combinator] Pomerado: Turn any website into an MCP for your agents, in minutes | Y Combinator, https://www.ycombinator.com/companies/pomerado
- [Pomerado, October 2026] Introduction - Pomerado, https://docs.pomerado.ai/
- [Aaron Epstein on X] Aaron Epstein on X, https://x.com/aaron_epstein/status/2114509061890998717
- [GitHub] GitHub - Pomerado/pomerado, https://github.com/Pomerado/pomerado
- [Company Launch Tracker, September 2026] Company Launch Tracker #110, https://companylaunchtracker.substack.com/p/company-launch-tracker-110
- [LinkedIn] Rudi MK - Porter | LinkedIn, https://www.linkedin.com/in/rudimk/