3 vertical AI platforms — one engineering team
dlaNGO is the first of three QA10 platforms — a grant application document generator for Polish foundations and associations. The algorithm matches calls to your statute, assembles the application package, you approve every paragraph. Pilot launches in July 2026 — in 2 months. Legal & Compliance AI and R&D Formulation follow in Q3 and Q4 2026.
/projekty/portfolio // three ventures, one infrastructure
Grant platform for foundations and associations. Hub of grant calls, application package generator, on-demand support.
See full description → 02 coming Q3 2026Polish legal order in an AI layer — no hallucinated articles. Contract review, GDPR audit, AI Act gap analysis.
See full description → 03 coming Q4 2026Predictive reformulation for cosmetics and supplements — stability, INCI, CPNP in one platform.
See full description →Stack under audit — not marketing. Every model decision documented, every retention log GDPR-compliant.
/projekty/problem // where it burns hardest
Reading regulations (PARP, NFOŚiGW, FIO, Active Citizens), matching the statute, budget in a spreadsheet, timeline, KPIs, final report. External consultants take 15–25% of the grant or 8,000 PLN flat. The board loses control over content — the application comes back as a PDF to sign.
Every amendment resets the team's knowledge base. There is no automation layer that understands the Polish legal system — generic LLMs hallucinate Civil Code articles.
Every raw material swap triggers a cycle of stability testing, compatibility checks, INCI documentation. R&D departments in SMEs don't have budget for predictive formulation models.
/projekty/dlango // QA10 flagship venture — pilot July 2026
dlaNGO combines a grant hub, document generator, and support layer in one workflow. The algorithm reads your statute, matches active calls (PARP, NFOŚiGW, FIO, NOWEFIO, PROO, Active Citizens, Erasmus+, Horizon Europe, local government grants, LOTTO, ORLEN, PKO, PZU) and assembles a complete package — application, budget, timeline, final report (WoP). No agencies. No loss of control.
In 2025 the Polish non-profit sector left an estimated 2.1 billion PLN on the table — grants no organization applied for. Not from lack of need. From lack of time, documents, knowledge of procedures, and money for grant consultants who take 15–25% of the grant value.
The average Polish foundation employs 3 people. The project coordinator spends 40–60% of their time on administrative work — reading regulations, matching the statute, stitching the budget in Excel. Weeks that can't be reclaimed. dlaNGO takes over that work and hands it back to you as a complete document package — with the right to approve or edit every sentence.
Hundreds of sources in one place. Filter matched to your statute, location, budget, and team competencies. Alerts in advance, calendar in your ecosystem.
Diagnostic survey calibrates the algorithm for a specific call. The package contains everything you submit to the grantor. You approve every paragraph — the algorithm suggests, you decide.
Legal and fundraising consultations in 30 / 60 min slots. Topical training for boards. Accounting service via external partner.
Excel + Dropbox doesn't read call regulations. Generic CRMs don't know FIO, NOWEFIO or PROO application sections. Grant agencies take a percentage of the grant and return a PDF for signing — you lose control over the project narrative, and the rationale for indicators reads identically to 12 other organizations served in parallel. dlaNGO keeps you in the decision-maker's seat — the algorithm reads the regulation, maps it onto your statute, drafts the first version, and you edit every paragraph or approve it wholesale. The application goes out in your name, in your voice, with your accountability — because you actually read it.
Work that previously took 3 weeks of consultant meetings closes in 60 minutes of survey plus 2 hours of editing. Cost: 0 PLN during pilot — in exchange, structural feedback that literally calibrates the product for your cohort.
/projekty/dlango/demo // live preview MVP
Below runs a slice of the dlaNGO engine — profile selection, matched calls, package generator, support layer. No login. No data storage. No signature required.
Wybierz typ Twojej organizacji. Dalej zobaczysz, co dlaNGO widzi w Twoim polu.
/projekty/dlango/segments // who it's for
You're establishing a foundation or association. We generate the statute, founding resolutions, KRS registration documents, first board regulations.
2–5 years of NGO, 3–8 projects per year. Alerts matched to your activity profile, full-cycle package generator.
Budget 500k PLN+, multi-project, multi-grantor. Predictive alerts 90 days before deadline, accounting integrations, management dashboard.
Vertical assistant for law firms and compliance departments. Trained on a corpus of KC, KSH, GDPR, AI Act, DSA — with source citation on every answer. Generic LLMs guess paragraphs. We show the source.
Platform for R&D departments in cosmetics, supplements, and household chemistry. The model predicts stability, raw material compatibility, and CPNP / REACH compliance before the first lab test. Fewer iterations. Fewer burned batches.
/projekty/roadmap // 18 months to DACH
July → September. 20 organizations, 12 weeks of feedback. L&C AI — alpha closed.
Own GPU layer (H200) — fine-tuning on Polish legal and grant corpus.
API integrations with 3 accounting systems (wFirma, iFirma, Optima). L&C AI — beta.
Entry into CZ, SK, RO — local support, local grant integrations.
/projekty/stack // why this won't fall over
Legal and narrative reasoning layer. Generating grant application sections, contract review, statute compliance analysis — where linguistic precision and auditability matter.
Multi-modal and predictive layer. PDF regulation parsing, financial statement data extraction, formulation stability prediction in R&D.
Own inference layer — H200 in Q4 2026. Fine-tuning vertical models on Polish legal and grant corpus — without sending data outside the EU.
ISO 27001 — audit Q4 2026. SOC 2 Type II — Q2 2027. GDPR — by design from day 1. LSO:ATOM architecture (details on /en/o-nas/) — full client data isolation at the object level.
/projekty/faq // recurring questions
The dlaNGO pilot starts in July 2026 — in 2 months. The first cohort is 20 organizations selected from applications. Cohort decision by 15 June 2026. You apply via the form, we run a 20-minute qualifying call, and sign a simple pilot collaboration memorandum. No right of first refusal, no exclusivity.
We index over 400 active funding sources in Poland: NOWEFIO and PROO (NIW — National Institute of Freedom), FIO, PARP, NFOŚiGW, Active Citizens — National Fund (EEA Grants), Erasmus+ and Horizon Europe (European Commission), CERV, local government (JST) grants, and corporate foundation programs — LOTTO, ORLEN, PKO, PZU. The algorithm matches each call to your statute, location, budget, and project history.
The algorithm reads your statute (sections on objectives, public benefit activities, decision-making bodies), your application history, KRS data, and location. It cross-references this with call regulations — priority sections, formal criteria, own-contribution requirements, the 2026 call schedule. Each matched call receives a relevance score (0–100%) with justification (which criteria you meet, which need supplementation).
The dlaNGO Generator works in 5 stages: diagnostic survey (12–18 questions), pulling the statute and KRS documents, drafting the application in the specific call's template (FIO has a different structure than NOWEFIO, NOWEFIO different than PROO), budget calculation with own-contribution under the regulation, timeline with measurable KPIs. Full package in 60 minutes — you edit or approve every paragraph before submission.
Yes — KGW and OSP are beneficiaries of programs like NOWEFIO (up to 250k PLN), JST grants, and ARiMR programs. dlaNGO handles their specifics — the KRKGW registry, no own-contribution requirement for projects under 20k PLN, simplified reporting. The pilot reserves 3 seats for KGW and 2 for OSP.
Own contribution is financial and non-financial resources brought by your organization. The Generator automatically calculates required own contribution for each call (NOWEFIO requires 25% for projects over 20k PLN, Active Citizens 10%, some grants 0%). Non-financial contribution (volunteering, rooms, equipment) is valued at GUS (Polish Statistics Office) rates — volunteer hours × hourly rate for that work category.
Active 12+ months, registered in KRS or KRKGW, with a history of 3+ submitted grant applications, board of 2+ persons. We favor mid-sized organizations (3–30 people in the team) that hit the Excel + Dropbox + WhatsApp wall. First 20 seats — foundations, associations, federations, advocacy organizations, KGW, OSP, local action groups.
0 PLN for 12 weeks of pilot (July → September 2026). In exchange — structural feedback bi-weekly and consent to an anonymized case study post-pilot. Three paths after the pilot: commercial subscription 1.0 with 40% alumni discount, opt-out with no obligation, or a seat on the advisory board (reserved for the 3 most engaged organizations).
Solely in the EU — AWS Frankfurt (eu-central-1) for the application layer, Google Vertex AI in region europe-west3 for PDF parsing. Bedrock and Vertex in zero-retention mode — your data never enters model training, never leaves the EEA. Beneficiary data is additionally pseudonymized at the application layer. Your organization is the controller (Art. 4(7) GDPR), QA10 is the processor; data processing agreement in the onboarding package.
CSV and XML export from day 1 of the pilot — for all popular accounting systems for NGOs. Full API integration with wFirma, iFirma, Comarch Optima, and Symfonia in Q1 2027 with the commercial 1.0 launch. Cost category, account, and counterparty sync — bidirectional.
After project completion, the Generator produces the Payment Application (WoP) in the grantor's template plus the substantive report with achieved indicators. It pulls data from project execution (attendance lists, invoices, volunteer lists, beneficiary surveys) and maps them 1:1 onto the report structure. Annual NGO reports and OPP (public benefit) reports — analogously.
Routing decision per task, not per vendor. Claude (Anthropic via AWS Bedrock) — legal reasoning, narrative, long regulations in a single pass (context window 200K/1M tokens). Gemini 3.1 Pro (Google Vertex) — multi-modal, PDF parsing, structured output for grant scoring. Vendor lock-in is an existential risk, not a product strategy.
It doesn't replace — it positions you so the consultant isn't needed for 80% of the work. In the 20% of hard cases (innovative Horizon Europe projects, cross-sector consortia, enterprise grants) the Support layer offers on-demand consultations with lawyers and fundraisers — 30 and 60 min slots, paid, but with no percentage of the grant.
The algorithm monitors grantor publications (NIW, PARP, NFOŚiGW, EC) 24/7. A detected regulation change triggers an alert to all organizations actively drafting under that call — with a concrete diff (which clauses, how they affect your application). Keeping the application aligned with the current regulation is our responsibility, not yours.
/projekty/contact // pilot July 2026 — cohort 01
We're opening the first dlaNGO pilot cohort in July 2026. We're looking for 20 NGOs that will submit real grant applications through the platform and return structural feedback. In exchange — 0 PLN for 12 weeks, 12 months of 1.0 with price lock-in, and an advisory board seat for the 3 most engaged.
Application is non-binding. Cohort decision by 15 June 2026. Pilot runs 1 July → 23 September 2026.