A revealing sentence in Abid Mohammed’s recent writing is not about artificial intelligence. It is about doubt. “Are you sure?” is the question that follows every elegant diagram once software meets a living business: its old databases, hurried users, peculiar approvals and rules nobody remembered to put in the manual. Mohammed has spent more than twenty years getting closer to that moment. First he wrote software. Then he advised clients, led product engineering and directed innovation. Now, as co-founder for technology at Cognida.ai, he works where the confident prototype meets the untidy institution.
This is unfashionable territory. Models attract fascination; integration attracts meetings. A demo can be completed by Friday. Long-term ownership comes with no such calendar. Yet Mohammed’s public record returns steadily to the prosaic things that decide whether technology earns a place in a company: boundaries, failure modes, data quality, security, controls and the people who will maintain the result.
He describes himself through four overlapping professions: architect, engineering leader, product manager and digital-solutions specialist. That list is less a collection of titles than a map of expanding responsibility. Every step takes him farther from the solitary certainty of code and closer to the negotiated reality of a system.
Before the model, there was the application
Mohammed studied computer science at S R T M University in Nanded from 1999 to 2003. His working life began the following year at e-Zest Solutions, where he was a software engineer. In 2006 he moved to Tata Consultancy Services as an associate software engineer. These were years of distributed and web applications, when “cloud” was not yet pasted onto every strategy slide and a software career still began, quite reasonably, with software.
In 2008 he joined Hitachi Consulting. The eight-year stretch that followed carried him through .NET consulting, offshore account management and senior management in consulting services. The combination matters. Consulting teaches a particular form of humility: a technically correct answer can still be commercially useless, operationally impossible or simply late. Account work adds another education. Someone has to translate between what a client asks for, what a team can build and what the budget will forgive.
From codebase to company
Software engineering at e-Zest and Tata Consultancy Services.
Consulting, account leadership and senior management at Hitachi Consulting.
Director roles in product engineering, innovation excellence and digital solutions at Hitachi Vantara.
Co-founder for technology at Cognida.ai.
Hitachi Vantara, where he worked from 2017 until 2022, widened the frame again. Across director-level roles in product engineering, innovation excellence and digital solutions, his work involved product development and digital transformation using cloud, analytics, connected devices and engineering. In 2018 he earned Google Cloud’s professional cloud architect certification. The credential is a small item in the chronology, but the word “architect” became the durable one.
A company built for the morning after the pilot
Mohammed became Cognida.ai’s co-founder for technology in April 2022. The company’s premise is plain enough to sound obvious: enterprises need AI that produces measurable outcomes and runs within the systems they already have. Achieving it is the complicated part. Cognida combines consulting and engineering with its Zunō accelerator platform, serving work across manufacturing, finance, technology and other industries.
The company announced a $15 million Series A led by Nexus Venture Partners in February 2025. At the time, Cognida said it had deployed solutions at more than 30 enterprises. Reported projects included speeding invoice processing, reducing customer churn and shrinking a manufacturer’s catalog-generation cycle from months to weeks. The numbers are company claims, but the pattern is instructive: each result is attached to a workflow, not to a model leaderboard.
That distinction suits Mohammed. On the Market Genius AI podcast, he discussed predictive models for churn, supply-chain optimization and assistants that automate business processes. The recurring advice was to select use cases carefully and set expectations before the technical enthusiasm outruns the organization. AI adoption, in this view, is less like buying a clever appliance and more like altering the plumbing while the building remains occupied.
In July 2025, Cognida introduced Codien, a tool that converts Selenium and Protractor test suites to Playwright and helps create new tests from plain-English instructions. Mohammed’s launch quote was characteristically functional: the team built it to help modernize testing “quickly and efficiently.” Codien also illustrates the broader philosophy. Its local-first architecture keeps source files on the user’s machine, while conversion and validation happen visibly. Speed is useful. Trust must be designed.
03 · Gravity winsWhy a prototype has no idea what awaits it
Mohammed has adopted a useful phrase from his co-founder Gopalakrishna Kuppuswamy: “Enterprise Gravity.” Small tools can move lightly. Serious systems encounter the pull of unique workflows, security requirements, legacy integrations, dirty data and ownership measured in years. An AI coding assistant can accelerate production, but acceleration does not repeal maintenance.
The forces a clean demo does not display
The proposed answer is “tribe coding”: a domain expert supplies context and judgment, an AI tool accelerates execution, and an engineer supplies structure, tests and architectural boundaries. It is not a theory that flatters lone geniuses. It says the quality of a system depends on people with different kinds of knowledge remaining in the room.
There is continuity here with Mohammed’s older writing. In 2017, he published articles about the questions teams should answer before choosing microservices and the organizational building blocks required for innovation. The technologies changed; the instinct survived. Architecture is a series of commitments. Fashion is a poor reason to make one.
Think first. Experiment sooner. Change your mind.
The temptation is to mistake caution for slowness. Mohammed’s more recent argument is nearly the reverse. AI contains too many new, fast-changing and non-deterministic parts for an architect to resolve everything on paper. Some decisions have to be learned through experiments. Begin with a hypothesis: perhaps a model can reason through the task, perhaps the data is rich enough, perhaps the workflow will accept the result. Then build only enough to find out.
Agents make those attempts cheaper. A team can compare alternatives and discard weak ideas before they become expensive commitments. Architecture consequently moves closer to the running system. Builders watch failures, revise assumptions and allow evidence to reshape the next version. Mohammed calls the process more fun, which may be a useful clue to his temperament. He is a skeptic, but a cheerful one. The purpose of doubt is not to stop the work. It is to give the work a chance.
His brief public comments on culture sound similar. In 2023, sharing a recruiting appeal, he wrote that Cognida was building a team to solve complex problems with innovative solutions “while having fun.” A former colleague’s recommendation describes him as technically capable, attentive to business and commercial objectives, and easy to work with. None of this yields a grand founder mythology. It offers something more useful: a picture of an engineering leader for whom rigor and sociability can occupy the same meeting.
Cognida’s scope continues to expand. In August 2026 it acquired Automate from Within, adding software for high-accountability accounting and operational workflows. Those are environments where an approximately correct answer may become a financial error, an audit problem or both. Mohammed’s response to the announcement emphasized evaluations, controls and auditability in production. It was another return to the morning after the demo.
There is a telling contrast between this work and the usual language of an AI launch. The announcement speaks about agents and platforms; the operating problem speaks about exceptions, approvals and evidence. A contract review system has to know when not to decide. An invoice workflow needs a path for the odd document that refuses to resemble the thousands before it. The human expert does not disappear. The expert moves to the uncertain edge, while software handles what can be made repeatable. That arrangement fits Mohammed’s larger pattern. He does not draw a border between technical and business concerns, then assign each side to a different room. The architecture includes both. A model, a policy, an integration and an accountable person become parts of the same design. If one is missing, the system is unfinished, however persuasive its answers may sound.
The arc of his career now looks unusually coherent. An engineer learns to make an application work. A consultant learns that working software can still fail its client. A product leader learns to balance repeatability with peculiar requirements. A founder learns that the product, team, customer and market form one larger system. AI makes that system faster and less predictable. It does not make judgment optional.
Mohammed’s aspiration, as expressed through his work rather than a slogan, is to narrow the distance between what AI promises and what an enterprise can responsibly operate. The task requires ambition, certainly. It also requires the unfashionable virtues: patience with old systems, interest in edge cases, respect for accountability and the confidence to throw away an attractive idea when the experiment says no.
A model can produce an answer in seconds. Making that answer belong inside a business may take years of accumulated context. Abid Mohammed has spent his career accumulating precisely that.