Use AI on your own business data. Nothing leaves your company.
NeoPhoenix gives you two things. Vectalk — ask questions about your contracts, policies, reports and records, and get answers with the page they came from. Neo Agents — AI workers that do the repetitive work overnight: invoice checks, vendor follow-ups, reports. Both run inside your own systems, so your data never goes to a public AI.
- Orion Logistics — Master Services AgreementExpires 30 Nov
If either party ends the agreement before the term, the exiting party pays a penalty of 12% of the remaining contract value.
Orion_MSA_2024.pdf · p. 14 - Apex Facility Services — Maintenance ContractExpires 15 Dec
Renewal notice must be given in writing; late notice attracts a penalty equal to two months of fees.
Apex_Maintenance_v3.pdf · p. 7 - Brightline Software — Licence AgreementExpires 31 Dec
Unused licences are not refundable, and cancellation within 60 days of expiry incurs a 5% penalty.
Brightline_Licence.pdf · p. 22
- 10 active Vectalk clients across manufacturing and D2C
- Pilot to production in 3 weeks, maximum
- ISO 27001 & ISO 42001 certified
- Works on your servers or your cloud
Every business wants to use AI. Most cannot use it on real data.
You have seen what ChatGPT can do. You have also seen the problem: the moment you paste a customer file, a contract or a salary sheet into it, that data has left your company. For a bank, a hospital or a listed company, that is against the law. For everyone else, it is still a risk nobody wants to explain to a customer.
Use public AI anyway — and hope nobody asks where the data went.
Hire a big consulting firm — and pay a price that was set for the top 150 companies in the country.
Wait — and watch competitors move first. (And from 13 May 2027, waiting also means fines under the DPDP Act.)
The AI tools were built for very large companies in other countries. The door was never built for you. So we built it.
Why 13 May 2027 mattersVectalk. ChatGPT, but on your own documents, inside your own infrastructure.
A compliance head asks, “Which vendor contracts expire this quarter with a penalty clause?” — and gets the three contracts and the exact clause, on her own screen, inside her own building. Nothing left.
Four things that are hard to copy:
- 01
Your data stays with you.
Deployed on your servers or your cloud. Nothing leaves.
- 02
Fixed rules decide compliance — not the AI.
Every determination is reproducible and traceable to the clause it came from.
- 03
Access is decided before the AI answers.
Roles, fields and masking are applied first; the model only sees what that person may see.
- 04
Every deployment faster and cheaper than the last.
Pilot to production in 3 weeks, maximum. Pilots start in the ₹6 lakh band, not the ₹6 crore one.
Opening the door is half the job. Someone still has to do the work.
Someone on your team is doing work nobody was hired to do.
Your accounts lead spends Monday matching invoices to purchase orders. Your operations head chases the same vendor documents every month. Your founder re-types the monthly report from four systems that do not talk to each other. None of it is in anyone's job description. All of it has to be done.
Neo Agents are AI workers built for your specific tasks. At 2 am an agent checks every invoice received that day against its contract, flags four, drafts the vendor emails and prepares the monthly report. At 9 am your operations head approves the four and sends the emails. Nothing left your systems.
Invoice and contract checks
Every invoice matched to its PO and contract overnight; exceptions on one screen at 9 am.
Vendor and customer documents
Missing documents requested, received and filed without a human chasing.
Reports that write themselves
The monthly report drafted from your systems; a person approves it, not assembles it.
Governed by design: every agent works under rules you set — who it may ask, what it may read, what needs a human sign-off. When the DPDP deadline arrives, you are already there.
Built, deployed and certified — before we ever ran an ad.
10 active Vectalk clients across manufacturing and D2C, with agents live in a white-label deployment for a client's end-customers.
ISO 27001:2022 and ISO 42001:2023 certified. SOC attestation in progress.
In conversation with government departments, a police department and regulated enterprises in banking, chemicals and energy.
- Hareen DesaiCEO & Co-FounderAI architect and product builder with a proven track record of engineering complex enterprise systems across highly regulated domains.
- Priyanka RavalCTO & Co-FounderDistributed systems engineer specializing in high-throughput AI platforms, multi-tenant cloud infrastructure, and secure enterprise architectures.
- Raunaq KatariaCOO & Founding MemberTechnology and business advisor guiding AI strategy, product positioning, and market direction.
- Shailee DesaiEngineering Manager & Founding MemberOversees engineering lifecycles, platform release velocity, and operational infrastructure.
AI your board can say yes to.
Thirty minutes. You bring one question your organisation cannot ask its data today, or one process nobody was hired to do. You leave with a one-page AI plan written for a board, not an IT team.
Book a 30-minute briefing