Founded in Pune in 2023.
50+ projects, 20+ clients, no bench.
Woodfrog builds data platforms, analytics, AI evaluation and internal applications for companies that already have the data and still cannot get a straight answer out of it. Every engagement is delivered by a pod of three experienced practitioners, and week one is a fixed-fee audit whose written output you keep whether or not you continue. This page is about how the firm works rather than what it believes.
- Founded
- 2023, Pune
- Delivered
- 50+ projects, 20+ clients
- People on your engagement
- 3, no bench
- Median to first production artefact
- 6 weeks
- Partner status
- Anthropic Build Partner
How an engagement runs
The same shape every time, priced and scoped before it starts.
- Week 1
Audit
A fixed fee, two calls and a written one-pager. The one-pager is yours whether or not you carry on.
- Weeks 2 to 3
Something real, in front of real users
The first working thing reaches the people who will actually use it. Narrow on purpose, and running on your data rather than in a sandbox.
- Weeks 4 to 6
Build
The rest of the first scope. Across 20+ companies, the median time to a first production-grade artefact is six weeks.
- Week 7 onward
Operate
Quarterly reviews and on-call governance. The people who built it are the people who pick up the phone.
The only part you commit to before you have written work in hand is the first week of six. Six weeks is the median time to a first production-grade artefact across 20+ companies.
What the Week 1 audit actually produces
Not a proposal dressed as a discovery, but a short document you can act on without us.
- 01
A fee agreed before anything starts
Week one is a fixed price. No time and materials, and no revised estimate at the end of it.
- 02
Call one, on what breaks
What you have, what goes wrong, who complains and how often.
- 03
The days in between
We look at the real tables, jobs and dashboards where you can give us access, and at the reports you send out where you cannot. The diagram and the system are usually different.
- 04
Call two, where we get argued with
We put a draft in front of your team before it is finished, so the people who know the business can tell us where we have read it wrong.
- 05
The one-pager
What we found, what we would build first, what it would cost, and what we would not build. Plain English, one page, no appendix of slides.
- 06
You keep it
Either way. If you take it to another firm and have it built cheaper, the week still did its job.
Four disciplines, staffed for depth
These are genuinely different jobs, and we would rather be good at four of them than available for forty.
- Data engineeringGetting data out of the systems that hold it and into a shape that can be joined, queried and trusted. Ingestion, modelling, orchestration, and the checks that stop bad rows becoming board slides.
- Analytics and BIFewer dashboards, agreed definitions, and numbers that reconcile across teams. This work now runs largely through Antvia, our own platform.
- AI evaluationDeciding whether a model or an agent is good enough to put in front of customers, with a test set and a scoring method rather than a demo and an opinion.
- Applications and automationThe software that replaces the spreadsheet somebody maintains at the weekend. Small internal applications and the automations around them.
How the firm is staffed
Practitioners per engagement
Delivery is a pod of three experienced people. Nobody is learning on your clock, and there is no handover in month two to a team you have never met.
People on the bench
We do not carry people between projects. A firm with idle staff has a standing incentive to put them on your work, and we would rather tell you we are full.
The people on the Week 1 call are the people you are still talking to when it is running in production.
What we publish, and what we will not
Four rules govern what appears on this site.
- 01
Figures are published after your Finance team has checked them
A saving or a revenue number goes on this site only once someone on the customer side who owns that number has confirmed it. Fewer case studies, and numbers that survive being asked about.
- 02
Every service page says where that service is the wrong answer
Our own manufacturing product puts it best on its site: “Saying this on the website costs a few signups and saves both of us a wasted month.” The same rule applies to the consulting pages.
- 03
Capabilities, never certifications we do not hold
Where a platform enforces column-level access control, PII masking or an audit trail, we describe exactly that. We do not present engineering controls as a compliance certification we have not been through.
- 04
NDA-friendly, fixed scope
Most of what we build is not something a client wants named. We work under NDA as a default, and scope is agreed and priced before the work starts rather than discovered during it.
Who we are not the right firm for
Three cases where the honest answer is to go elsewhere.
You want a large body shop
If the requirement is twenty people on site next month, we are not it. We field three per engagement and there is no bench behind them to scale up from.
You already have the answer and want it confirmed
Week one exists to test the premise. If the audit disagrees with the plan you walked in with, that disagreement is what gets written down.
Nobody on your side can sign off a metric definition
Somebody has to be able to say what an active customer is, and be listened to when they say it. Without that person the build stalls where two systems disagree and no one can break the tie.
The second and third of these can usually be fixed by a conversation with the right person in the room. The first one cannot.
Two products, for the same reason
The same problems kept arriving, and rebuilding the answer from scratch on every engagement stopped being defensible.
- AntviaA production-grade lakehouse, already built, deployed into your own cloud. Bronze to silver to gold, 300+ supported sources, column-level access control and PII masking enforced on every query. Roughly a quarter of the cost of building it yourself.
- UloborusBuying, receiving, quality, stock, making, selling, dispatch and invoicing on one thread for Indian process manufacturers, with GST paperwork generated from the same records the plant runs on. Built module by module with Rajaram Consumer Care in Sangli.
- Why we ship products at allA consultancy that rebuilds the first six weeks of a data platform on every engagement is charging clients for its own repetition. Building it once and deploying it into the customer’s cloud is the more honest version.
What has happened so far
The firm in the order it actually happened.
- 2023
Founded in Pune
Founded in Pune as a data engineering and analytics practice.
- 2023
First enterprise client onboarded
The engagement model, a fixed first week and a pod of three, came out of that first delivery rather than out of a strategy document.
- 2024
AI agent development added
Agent work joined the practice, and with it a view that evaluation, not model choice, is where agent work usually gets decided.
- 2025
Clients across industries
Fintech, B2B SaaS, retail and public sector.
- Now
Anthropic Build Partner
Shipping production agents on Claude, and in the Claude Partner Network from launch as a Registered partner with a path to Select. 50+ projects delivered for 20+ clients.
The people who would be on your engagement
Not a leadership page: the pod on your work is drawn from these practitioners.
Roles, one-liners and LinkedIn profiles are on the team page.
Questions we get asked
How big is Woodfrog?
Small, and deliberately. We publish what we have delivered, 50+ projects for 20+ clients since 2023, rather than a headcount, because headcount is the number consultancies inflate. Three experienced people are on your engagement, and they are not shared out across five others.
Who actually does the work?
The pod of three. Nobody is learning on your clock, and there is no handover to a team you have never met. Whoever runs your Week 1 calls is on the build.
What happens if the audit says do nothing?
It gets written in the one-pager and you keep it. A firm that cannot reach that conclusion is not really auditing anything.
Do you work under NDA?
Yes, as a default. Most of our work is not named publicly. Where a case study exists, it is published because the client agreed to it and their Finance team checked the numbers in it.
Are Antvia and Uloborus resold software?
No. Both are built and run by Woodfrog engineers. Antvia deploys into your own cloud on open formats, so you can export or take it over. Uloborus is a working product built alongside a live factory.
Is anything in Uloborus analytical today?
Yes. Every list view in it is analytical from day one, with a stats strip, query search and a breakdown panel, and that is genuinely shipped.
Does Uloborus ship AI agents today?
No. Anomaly and margin-leak detection are on the roadmap and not in the product. What Uloborus does today is produce the clean, governed operational dataset that agents can later be built on.
Where are you and who do you work with?
Pune, India, delivering across India, APAC and the US. NDA-friendly, fixed scope.
If this is the way you want to work
The cheapest way to find out whether we are useful to you is the fixed-fee first week: two calls, a written one-pager, and a document you keep and can act on with or without us. If you would rather ask a question first, hello@woodfrog.tech reaches the people who do the work. Pune, India, delivering across India, APAC and the US. Data and AI engineering for modern businesses.