Everything a chatbox skips. Built in.
SaySo OS is the layer between you and every LLM. Below is the full breakdown — what each capability is, why it matters, and how to use it. New accounts start with 300 free credits (≈ 60–150 runs) plus 20 free every day.
Prompt engineering, done for you
You describe the task in plain words. SaySo turns it into a structured, model-ready prompt — automatically.
Framework intelligence
SaySo reads your task and auto-selects a proven prompt structure — COSTAR, RCIT, CRISPE and more.
- why
- The framework is what separates a vague reply from a precise one. Most people never learn them; you don't have to.
- how
- Just type. The detected framework appears as a chip — keep it, or switch with one click. Each framework exposes the exact fields it needs.
Smart fields & guardrails
Role, audience, tone, format, constraints — broken into clear slots with examples and inline guidance.
- why
- Every field you fill removes ambiguity, so the model stops guessing and starts delivering what you meant.
- how
- Fill what's relevant, skip the rest. Hover any field for a hint. Leave it blank and SaySo still sets a sensible default.
Output playbooks
Picking an output type (LinkedIn post, cold email, PRD, code review…) injects expert structure for that exact format.
- why
- A LinkedIn post and a legal memo need totally different shapes. Playbooks teach the model what 'good' looks like for each.
- how
- Choose your output format in the composer — the matching playbook is added behind the scenes. Your own instructions always win.
Run it your way
One cockpit for every model, every depth, every budget.
Multi-model, one cockpit
Run GPT, Gemini, Claude and more from a single surface — single-shot or side-by-side compare.
- why
- No model is best at everything. Comparing answers shows you which one nails your task before you commit.
- how
- Pick a model for single mode, or select 2+ and hit run to compare. Each lane is tinted in that model's brand colour.
Quick · Standard · Detailed
A response-mode switch that tells the model how much depth you want.
- why
- Sometimes you want a one-line answer; sometimes you want the full breakdown. One toggle, no re-prompting.
- how
- Use the mode dropdown in the composer row. Quick keeps it tight, Detailed adds reasoning and ancillary help. Standard is default.
Thinking modes
Chain-of-thought, tree-of-thought, ReAct and more — switched on with a tap.
- why
- Harder problems need structured reasoning. These techniques are buried in research papers; here they're a toggle.
- how
- Open Advanced → choose a thinking method. SaySo wires the reasoning scaffold into your prompt automatically.
Pre-flight cost in ₹
See tokens, credits and the exact rupee cost before you send — and per lane after.
- why
- No surprise bills, no guessing. You always know what a run costs before spending a credit.
- how
- The cost chip updates live as you build. Set per-run caps in Cost controls to stay on budget.
Context, scale & reuse
Bring your own material, break down big asks, and keep your best setups.
Reference uploads
Attach your own docs, notes and files as context for the answer.
- why
- The model works from your material instead of generic knowledge — used as reference, never a cage.
- how
- Click Attach in the composer. Files become chips and are merged into the run payload, not your visible text.
Break big asks down
Hand over a heavy task; SaySo splits it into a plan, executes step by step, and merges one answer.
- why
- Large requests overwhelm a single prompt. Decomposition keeps quality high across long, multi-part work.
- how
- Just ask for the big thing. When a task is complex, SaySo shows the plan it's working through above the result.
Templates & memory
Save your best setups as reusable templates that show up in your Templates library.
- why
- Stop rebuilding the same prompt. Your go-to setups become one-click starting points.
- how
- Hit the Save (template) icon by Run. Saved setups appear on the Templates page, ready to load and tweak.
Confidence & output
Know it's right before you spend, then take it anywhere.
Quality score + mismatch alerts
A live quality read and warnings flag gaps or conflicting choices before you run.
- why
- Catch a weak or contradictory prompt early — before it costs you a credit and a bad answer.
- how
- Watch the quality meter as you build. Mismatch warnings appear inline with a one-tap fix suggestion.
AI clarifying questions
SaySo can ask you 1–4 targeted questions — multiple choice plus write-your-own — to sharpen the brief.
- why
- A couple of quick answers often turns a decent result into exactly what you pictured.
- how
- Tap Clarify in the composer. Answer the cards, and your picks fold neatly into the prompt.
Export anything, any way
Copy or download every result — text as TXT/MD/HTML/DOC/JSON, images as PNG/JPG/WEBP, plus Print/PDF.
- why
- Your output should live wherever you work, in the format you actually need.
- how
- Use the copy/export menu under each answer. Image results download as real image files, not prompt text.
Live coach
A handholding coach above the composer that adapts to whatever panel you're using right now.
- why
- Learn the why behind every control as you go — frameworks, advanced settings, reviews, personalisation.
- how
- It updates as you move around the Control Room. Dismiss it anytime with ×, bring it back with one tap.
heads up — AI is a brilliant co-pilot, not an oracle. It can draft genius in seconds and also invent a fact with a straight face. SaySo makes it sharper than raw chat, but the final call is always yours — give every answer a once-over before you ship it. 🫡