AIYING ARTIFACT #2
The Website-to-Proposal Engine
Turn a public website into an evidence-based opportunity, tailored proposal and credible first communication.
AIYING — Accomplishing with AI | TAJ — Think → Ask → Judge
Don’t ask AI to “write a proposal” from a website. Build an evidence trail from the organisation’s reality to a proposal it can reasonably consider.
Executive Summary
An organisation’s website contains far more than descriptive information. It may reveal its priorities, programmes, audiences, investments, challenges, language, partnerships and direction. Yet copying a URL into an AI tool and asking for a proposal rarely produces something genuinely useful. The result may sound professional while remaining generic, speculative or disconnected from what the organisation actually needs.
The Website-to-Proposal Engine is a reusable AIYING artifact for moving from public organisational evidence to a decision-ready proposal. It combines web research, human interpretation, AI-assisted production and TAJ verification. It makes a vital distinction between what a source states, what we may reasonably infer, what remains an assumption, and what must be verified directly.
The engine produces five connected outputs:
- An evidence map of the organisation
- A ranked opportunity hypothesis
- A matched and bounded offering
- A tailored proposal
- A personalised first communication
Its purpose is not to automate persuasion. Its purpose is to improve relevance, truthfulness, speed and human judgment.
Core flow
Website → Evidence → Organisational pattern → Opportunity hypothesis → Offering match → Proposal → TAJ verification → Human outreach → Response → Learning
The Problem: Websites Are Usually Read, Not Converted into Work
Suppose you want to approach a university, company, hospital, training organisation, MSME or nonprofit. You visit its website. You read the About page, programmes, news, leadership messages and recent initiatives. You may understand the organisation better—but the information does not automatically become a useful action.
Three common shortcuts then appear:
- The generic proposal: The same offering is sent to everyone with only the name changed.
- The overconfident inference: A public signal is treated as proof of an internal need.
- The impressive AI draft: Fluent language conceals weak evidence, invented facts or poor institutional fit.
The problem is not a lack of information or prompting skill. The missing element is a disciplined bridge between evidence and action.
Idea, Prompt and Artifact Are Not the Same
| Form |
What it provides |
What is missing |
| Idea |
“Use an organisation’s website to create a tailored proposal.” |
Method, controls, verification and reusable structure |
| Prompt |
An instruction asking AI to research and draft |
A complete workflow and human decision gates |
| Artifact |
A repeatable engine with inputs, stages, outputs, checks and reuse logic |
It still requires responsible human use |
A prompt may be part of an artifact. It is not the artifact by itself. The Website-to-Proposal Engine includes a power prompt, but it also tells the user what evidence to collect, what decisions to retain, how to verify the result and how to learn from the response.
Who Is This Artifact For?
The engine is intended for people who must understand an organisation before offering something meaningful:
- Consultants and independent professionals
- Training and education providers
- Institutional partnership teams
- Business-development and sales professionals
- Faculty-development specialists
- MSME solution providers
- Nonprofits seeking institutional collaboration
- Researchers developing partnership concepts
- Jobseekers preparing a value proposition for a target employer
It is most useful when the offering must be adapted to the recipient rather than merely advertised.
The Complete Website-to-Proposal Engine
Stage 1 — Define the Outcome
Do not begin with “study this website.” Begin by specifying what real-world decision or action the work should support.
Complete this sentence:
I want to understand [organisation] well enough to propose [type of value] for [intended users], leading to [desired next decision].
Example: “I want to understand a university well enough to propose a practical faculty AI intervention, leading to a pilot-workshop discussion.”
Output: One outcome statement and one intended next decision.
Stage 2 — Establish the Evidence Boundary
Identify the sources that can legitimately support the work. The organisation’s own website is the primary source, but relevant evidence may also include official reports, programme pages, prospectuses, accreditation documents, annual reports, verified news releases, leadership profiles and official recruitment pages.
Record every important observation in four columns:
| Source |
Evidence |
Date checked |
Relevance |
| Official page |
What the page explicitly states |
Current date |
Why it matters to the proposed value |
Rule: Never turn promotional language into an independent fact without qualification. Attribute institutional claims to the institution when appropriate.
Output: A source-linked evidence map.
Stage 3 — Read the Organisational Pattern
A single fact rarely reveals an opportunity. Look for patterns across multiple signals:
- Direction: What future is the organisation moving toward?
- Investment: Where is it committing infrastructure, programmes or partnerships?
- Audience: Who is expected to benefit?
- Activity: What is already being implemented?
- Gap: What capability may be required to convert investment into practice?
- Language: How does the organisation describe its own aspirations?
This is Pattern AI in practice: not merely collecting facts, but finding a recognizable and actionable relationship among them.
Output: A short organisational pattern statement.
Stage 4 — Separate Fact, Inference, Assumption and Unknown
This is the engine’s most important truth-control mechanism.
| Category |
Meaning |
Permitted use |
| Fact |
Explicitly supported by a reliable source |
May be stated with citation or attribution |
| Inference |
A reasonable interpretation of several facts |
Label as an interpretation or opportunity |
| Assumption |
A proposition not yet supported |
Test or remove; do not present as fact |
| Unknown |
Information unavailable publicly |
Convert into a discovery question |
Output: A claim ledger that prevents hallucination and overclaiming.
Stage 5 — Generate and Rank Opportunity Hypotheses
Generate three to five possible opportunities before choosing one. Rank them using five questions:
- Is it supported by evidence?
- Is it relevant to the organisation’s direction?
- Does it complement rather than duplicate existing work?
- Can the proposed provider credibly deliver it?
- Can it lead to a realistic next step?
A simple score from 1 to 5 for each question can expose attractive but weakly supported ideas. The highest total should not be accepted automatically; human judgment must still consider ethics, timing, relationships and strategic fit.
Output: One primary opportunity and up to two alternatives.
Stage 6 — Match the Offering
Do not force the organisation into a predetermined service. Adapt the offering only within your actual capability.
Build the match explicitly:
- Observed signal: What did the organisation reveal?
- Possible need: What capability might strengthen that direction?
- Relevant offering: What can you genuinely provide?
- Expected outcome: What could participants or the organisation produce?
- Boundary: What will the offering not claim or solve?
Output: A one-paragraph evidence-to-value proposition.
Stage 7 — Design the Proposal Architecture
A decision-ready proposal should answer the recipient’s practical questions in a natural sequence:
- Why this proposal, for this organisation, now?
- What problem or opportunity does it address?
- Who is it for?
- What exactly will happen?
- What will participants or the organisation produce?
- How will quality and human control be maintained?
- What resources, time and participation are required?
- What does it cost?
- What is the smallest sensible next step?
The proposal should use the organisation’s language where accurate, but it should never imitate endorsement or imply a relationship that does not yet exist.
Output: A tailored proposal ready for human review.
Stage 8 — Apply the TAJ Verification Gate
Before release, stop production and switch to judgment.
THINK
- Is the intended outcome still clear?
- Does the proposal serve the recipient as well as the sender?
- Are consequential decisions still human-owned?
ASK
- Were the right sources examined?
- Were contrary or limiting signals considered?
- Were assumptions converted into questions where possible?
JUDGE
- Can every institutional claim be traced to evidence?
- Are facts, inferences and assumptions clearly separated?
- Is the offering relevant rather than generic?
- Are outcomes concrete and feasible?
- Are scope, requirements and commercial terms internally consistent?
- Has sensitive, private or unnecessary information been excluded?
- Would a responsible professional be comfortable signing and sending it?
Output: Approved, revise or stop.
Stage 9 — Create the First Communication
The message should not repeat the entire proposal. It should explain why the recipient is receiving it and make the next action easy.
A useful structure is:
- Personal and contextual opening
- Name of the attached proposal
- One-sentence relevance statement
- One-sentence description of the practical outcome
- A respectful request for review, guidance or discussion
Output: An email, LinkedIn message, letter or phone script appropriate to the relationship.
Stage 10 — Capture Response and Convert Learning into Reuse
The engine does not end when the proposal is sent. Record:
- Recipient response
- Questions or objections
- Which evidence or value proposition attracted attention
- What was misunderstood
- What required customisation
- Whether the opportunity advanced, paused or stopped
- What should change before the next use
Only after repeated successful applications should stable parts become a template, skill, scheduled workflow or operating system.
Output: A field note and an improved version of the engine.
The Power Prompt
Role: Act as an evidence-led proposal architect working under the AIYING framework and TAJ: Think → Ask → Judge.
Organisation: [NAME]
Official website: [URL]
My offering or capabilities: [DESCRIBE]
Intended users: [AUDIENCE]
Desired next decision: [MEETING / PILOT / REVIEW / PARTNERSHIP DISCUSSION]
Constraints: [TIME, COST, GEOGRAPHY, DELIVERY FORMAT, EXCLUSIONS]
Complete the work in the following order:
1. Restate the intended outcome and identify any critical missing input.
2. Examine relevant official pages and other reliable, current sources.
3. Create an evidence map with source, date, explicit evidence and relevance.
4. Identify the organisation’s direction, investments, audiences, current initiatives and recurring language.
5. Maintain a claim ledger containing four categories: verified fact, reasonable inference, unverified assumption and unknown.
6. Generate three opportunity hypotheses. Rank them for evidence, relevance, complementarity, provider fit and feasibility.
7. Recommend one opportunity, explain why it is strongest and state its limitations.
8. Map my genuine capabilities to that opportunity without inventing experience, outcomes, relationships or endorsements.
9. Draft a tailored proposal containing rationale, intended users, objectives, activities, outputs, quality controls, requirements, commercial terms if supplied, boundaries and next step.
10. Draft a concise first communication appropriate to the specified relationship and channel.
11. Run a TAJ verification audit. Flag every unsupported claim, hidden assumption, stale fact, contradiction, privacy concern and unrealistic promise.
12. Stop for my approval before treating the proposal as final or sending anything.
Do not infer a confirmed internal need from public website evidence. Present opportunity hypotheses as hypotheses until validated by the organisation. Use citations or source links for material external claims.
Important Disclosure
This is an independent AIYING field application based on publicly available information. Dayananda Sagar University has not sponsored, endorsed or validated this artifact.
Worked Example: Dayananda Sagar University
The first field application began with the website of Dayananda Sagar University (DSU). The goal was to move beyond a website summary and identify a credible faculty-development opportunity.
Verified evidence
DSU publicly positions itself as an AI-first university. Its official pages describe an AI-first learning model, an Academic AI Factory, NVIDIA DGX Blackwell infrastructure and programmes that expose learners to modern AI development and deployment workflows. Its leadership message also emphasises education, research, innovation, entrepreneurship and multidisciplinary experiential learning.
Official sources consulted:
Reasonable inference
An institution with visible AI programmes and infrastructure may benefit more from a common faculty work protocol than from another general introduction to AI. This was an opportunity hypothesis—not a claim that DSU had officially declared such a gap.
Offering match
AIYING could contribute a practical cross-school protocol through which faculty bring authentic teaching, research, assessment or administrative work; collaborate with AI; judge the output; and convert useful methods into reusable practice.
Artifact produced
The process produced a tailored proposal titled:
From AI-First University to AI-First Faculty Practice
A Three-Hour Hands-On AIYING Faculty Work Protocol Lab
The proposed working sequence was:
Outcome → Route → Collaborate → Produce → Judge → Learn → Reuse → Scale
The proposal was reviewed and shared with the university’s Vice Chancellor. At the time of this publication, a response and institutional outcome should not be assumed. That pending status is part of honest verification.
A Compact One-Page Engine
| Stage |
Question |
Output |
| 1. Outcome | What decision should this support? | Outcome statement |
| 2. Evidence | What do reliable sources explicitly show? | Evidence map |
| 3. Pattern | What direction appears across the signals? | Pattern statement |
| 4. Claims | What is fact, inference, assumption or unknown? | Claim ledger |
| 5. Opportunity | Which hypothesis has the strongest fit? | Ranked opportunities |
| 6. Match | What can we credibly contribute? | Value proposition |
| 7. Proposal | What must the recipient decide? | Decision-ready proposal |
| 8. Verify | Is it true, relevant and feasible? | TAJ decision |
| 9. Communicate | Why should this person review it? | First message |
| 10. Learn | What did the response teach us? | Field note and revision |
TAJ Verification Checklist
THINK
- ☐ The real-world outcome is explicit.
- ☐ The intended users and decision-maker are identified.
- ☐ The proposal creates value for the recipient, not only the sender.
- ☐ Human ownership of consequential decisions is clear.
ASK
- ☐ Current, relevant and preferably primary sources were used.
- ☐ Important sources and dates are recorded.
- ☐ The AI was instructed to preserve uncertainty.
- ☐ Missing internal information became discovery questions.
- ☐ No private data was collected unnecessarily.
JUDGE
- ☐ Every material fact is traceable.
- ☐ Facts, inferences, assumptions and unknowns are separated.
- ☐ The proposal does not imply endorsement or confirmed need.
- ☐ The offering matches demonstrated capability.
- ☐ Outcomes, duration, requirements and fees are consistent.
- ☐ The language is specific, respectful and free of empty flattery.
- ☐ The smallest sensible next step is clear.
- ☐ A human has reviewed and approved the final communication.
Suggested Applications
The engine can be adapted well beyond faculty-development proposals:
- Education: curriculum workshops, student programmes, employability interventions, research partnerships and institutional consulting
- Enterprise: capability programmes, AI adoption, workforce development and innovation workshops
- Training companies: evidence-led prospecting and course matching
- Healthcare: programme-aligned AI literacy and professional-development proposals
- MSMEs: operational improvement, digital transformation and targeted AI adoption
- Jobseekers: employer-specific contribution proposals instead of generic applications
- Researchers: collaboration concepts grounded in institutional priorities and facilities
- Nonprofits: partnership proposals linked to published missions and programmes
When Not to Use It
Do not use the engine to fabricate personalisation at scale, scrape restricted information, infer sensitive attributes, impersonate an insider, conceal automated outreach or present speculation as organisational knowledge. It is also unsuitable when the decision requires confidential internal data that you do not possess.
In such cases, the correct output may be a discovery request—not a proposal.
From One-Time Work to a Reusable System
A single successful proposal does not prove a universal engine. The responsible path is progressive:
- Apply the method to one authentic case.
- Record what was evidence, judgment and production.
- Test it with a second, different organisation.
- Compare which parts remained stable and which required adaptation.
- Convert only the stable components into templates or skills.
- Keep relationship decisions and final approval human-controlled.
The DSU proposal is the first field application. The next validation should use an organisation from a different context—perhaps an enterprise, healthcare institution, training company or MSME. That contrast will reveal whether the engine is truly universal or still education-shaped.
Conclusion: The Website Is an Entry Point, Not an Answer
A website can tell us what an organisation chooses to make public. AI can help us find, organise and connect those signals. But neither the website nor AI can authoritatively tell us the organisation’s complete internal need.
The Website-to-Proposal Engine therefore works through disciplined progression:
Observe carefully. Infer cautiously. Offer credibly. Verify rigorously. Communicate humanly. Learn continuously.
That is how a URL becomes more than a source of information. It becomes the beginning of meaningful, evidence-led work.
