Critical Organ Care · Tech Bets 2036

A programme for Vantive India R&D · Bangalore

August 2026

The promise

What you asked for

Place Bangalore's bets.

Not breakthroughs.

The programme's job, in one sentence

Surface 20–30 candidate technology bets
Set most of them down early and gladly, on evidence
Hand you 2–3 that held up

The one structural idea

Put the forecast before the problem

Method from FORMAT — EU FP7 technology forecasting programme

What a team brings back first

A cited state-of-the-art landscape
A 10-year forecast for its sub-domain
A white-space map — where the field is open

The arc and the two gates

Two gates, and what happens at each

Gate 1 · Choose	~30 bets in · 4–5 carried forward
Gate 2 · Commit	2–3 backed · named sponsor · budget line

The most important hour in the programme

You write the choosing criteria

before you see a single idea.

Two ways to run it

Same method, different commitment

Both: 10 teams × 5 people	Foresight Studio	Venture Studio
In one line	Find the bets	Find them and fund them
Duration	7–8 months	11 months
Landscape · forecast · two gates	✓	✓
Capital committed at Gate 1	—	✓
Teams bid for their own ground	—	✓
Panel of critical friends	—	✓
Forecast Cell — capability stays	—	✓

One question decides which

Does Vantive already have

a working route to fund a concept

that is not yet on a product roadmap?

What you get, and what it costs

What exists at the end — either way

2–3 bets that held up under honest testing
Invention disclosures ready for counsel
A 10-year roadmap the CTO office owns

Two options, two numbers

10 teams · 50 people	Foresight Studio	Venture Studio
Duration	7–8 months	11 months
Consulting & training	12 days · ₹15,00,000	19 days · ₹23,75,000
Software licenses	₹9,00,000	₹9,00,000
Total, excluding GST	₹24,00,000	₹32,75,000
GST at 18%, charged separately	₹4,32,000	₹5,89,500

Two lines and two invoices: consulting from me personally, the platform from Spirelia. The software line is ten team licences — one per team, up to six named users each, which is the five team members plus a floating slot for a mentor, and a free observer seat for the innovation office. It is licensed per team rather than per seat, which is unusual: comparable landscape and patent-analytics platforms run ten to thirty thousand dollars per seat per year. The year-one price carries a founding-customer discount of fifty per cent, and I would rather be direct that what we ask in return is a named reference and a written case study. Where the software runs is a separate menu — shared, dedicated, or entirely inside your own cloud — and I would start at the simplest and let your security review pull us up it.

A disclosure, before you weigh any of it

Utkarsha and Garuda are products of Spirelia, of which I am a founder and director.

They are here because the gate criteria require artefacts they produce.

You are free to substitute an equivalent platform — I will support the integration at no change in fee.

What we need from you

Three decisions

Studio or Venture Studio — the funding-route question
Team count and sub-domains — 10 teams, 3–5 domains
A named sponsor who will write the choosing criteria

Thank you

Place the bets you can defend.

Backup · the intrapreneurship frame

The Venture Studio is an intrapreneurship programme

Teams bid for their ground — nobody is handed a challenge
Named inventors on every disclosure
Capital moves in the room, in front of everyone

Mindtree, and what made it work

Element	In this programme
Business goals	10-year roadmap · a funded bet
Employee motivation	bid for your ground · named inventors
Innovation	systematic, from evidence
Brand building	a centre that originates

I ran an intrapreneurship study of Mindtree some years ago, and four elements separated it from the programmes that fade. Real business goals, so the work had somewhere to land. Motivation that let people pursue their own idea rather than an assigned one. Innovation that was systematic. And a brand story the outside world could hear. Three map onto this programme directly. The fourth is where I differ deliberately — Mindtree crowd-sourced its ideas and I start from evidence instead, for the reason we discussed. Everything else transfers, including the part that matters most, which is that people worked on what they chose. And there is an outside-facing version: a newly independent, listed company producing two or three evidenced ten-year bets out of Bangalore is a story analysts can hear. It is also a retention instrument, because the engineers you most want to keep are the ones who leave when there is no route for an idea that is not on this year's plan.

Backup · the AI track

At least one sub-domain must be AI-enabled

Closed-loop adaptive therapy control
Predicting circuit and filter clotting before it happens
A patient digital twin — test the prescription before you run it

The part that usually stalls medtech AI

FDA PCCP — how a learning model stays cleared
EU AI Act high-risk, stacked on MDR
IEC 62304 — and the data-ownership question

Most innovation programmes produce AI concepts that stall in regulatory review eighteen months later, and that is avoidable. So on this track the business person on the team owns the clearance path from Phase 2 onward, as a real job rather than a token seat. A predetermined change control plan is how a model that keeps learning stays cleared without a fresh submission each time. The EU AI Act's high-risk classification lands on medical device AI and stacks on top of MDR. And underneath both sits the question that decides these projects: what data exists, who owns it, is it labelled, does it leave the device, and does the DPA permit training on it. Alongside all this, teams also mine the FDA adverse-event and recall databases — yours and the competition's — which is public, systematic, and where genuine unmet need hides in the words of the people it happened to.

Backup · the platform

Why software is in this proposal at all

The Phase-1 gate artefact

cannot be produced by hand

in five weeks.

Your model, your tenant

Azure OpenAI in your own tenant — recommended
Or your AWS account · or fully local
Prompts never leave your boundary · usage not metered

This is what your security review will care about. The queries teams type are the roadmap — where is the white space in extracorporeal organ support — and you should not be relaxed about where they go. The tools run against your own model deployment. If you have Azure OpenAI in your tenant, pointing our platform at it is a configuration change, and your prompts never leave your Azure subscription; it also draws on committed Azure spend you have to consume anyway. Your own AWS account works the same way, and fully local is supported. Under any of these we do not meter your usage, because it is your inference. One question I would like answered in the first call rather than the first workshop: do you have Azure OpenAI provisioned, and is there headroom? If not, provisioning inside a regulated enterprise runs three to five weeks, and that can run parallel to onboarding only if we start it early.

One design rule worth your attention

The model that checks a citation

is never the model that wrote it.

Backup · after the gate

Make passing a gate mean something

Teams carried forward keep their licence 12 months past programme end
Gate-2 teams get Forecast Cell seats
The Clarity Award — for a team that sets its own bet down, with evidence

If a bet becomes a venture

Every bet backed at Gate 2 carries

12 months of platform licensing

into its venture life, free.

Backup · likely questions

"Why not start from our challenge list?"

Your list maps what you already know hurts
Those problems are known, catalogued, mostly incremental
We will use it — as a cross-check at Gate 1, not as the input

"What if none of them hold up?"

You still have a cited landscape and a 10-year roadmap
And you learned it in 8 months, not three years
Setting bets down early is the deliverable

"Who owns the IP?"

Vantive, entirely — inventions, disclosures, roadmap, data
Spirelia owns only its software
No training on your prompts or your corpora

"What does this cost us in people-time?"

~*4 hours a week* per participant, 50 people
Leadership: two gate days plus Phase 0
Concentrated in Phase 1 and the proof sprint

"Can we run 14 teams?"

The design holds; the licence cap is 10 teams
Beyond 10 is a pricing conversation — before the proposal
Coaching time scales with team count too