Documentation
AIGENCY Documentation
Everything you need to use the platform features and integrate via API.
Quick Start
AIGENCY lets you chat with AI assistants. After creating an account and entering the dashboard, you pick an assistant and start sending messages. Everything runs in the browser; there is nothing to install.
Create your account
Sign up with your email and click the verification link. You receive a welcome credit.
Choose an assistant
Click 'New Chat' in the left menu to see the assistant list. Alparslan is general purpose; Tural is for law, Altay for health, Algın for entrepreneurship, Umay for relationships and communication.
Send a message
Type in the box at the bottom and press Enter. The assistant's response streams word by word.
Your chat is saved
Your conversations are kept in the left menu; when you reopen them, you continue where you left off.
Assistants
Each assistant in AIGENCY is trained on its own area of expertise; they greet you with different personalities and working styles. Asking the same problem to different assistants can reveal different approaches. To switch assistants, just open a new chat with another one from the left menu; your existing chats stay intact.
Alparslan
Software & EngineeringGeneral-purpose assistant trained as an AI engineer. Expert in writing code, architectural design, data science, mathematics, and technical problem solving.
Tural
Legal AdvisoryLawyer profile. Searches high-court rulings of Türkiye (Court of Cassation, Constitutional Court, Council of State); maintains and expands your case file across turns, finds precedents, suggests legal reasoning.
Altay
HealthHealth data analysis and clinical decision support. Asks about your symptoms and discusses possible causes; does not replace a physician's diagnosis, presents structured information to be evaluated together with your doctor.
Algın
EntrepreneurshipBuilt on entrepreneurship principles. Advises on idea validation, business model, customer segments, go-to-market, financial projections, and investor pitches.
Umay
Relationships & CommunicationGuidance on relationship dynamics, communication skills, trust-building, and therapeutic topics. Offers practical suggestions to make your relationships more fulfilling and meaningful.
Göbeklitepe AI
History & ArchaeologyTalks about Göbeklitepe and other ancient civilizations of Anatolia, archaeology, and antique cultures; explains your questions with full historical context.
Udagan
Turkish Coffee FortuneExpert in Turkish coffee fortune telling and mystical interpretation. Describe your cup; receive unique readings based on symbols, directions, and traditional interpretive patterns.
Agent Mode
You will see a green button labeled "Agent" in the chat header. When enabled, the assistant performs web searches, cites sources, and can fetch images/videos. When disabled, it is classic chat mode (fast, answers only from what it already knows).
On
- · Performs up-to-date research
- · Responds in 5-10 seconds
- · Web sources, images
- · Slightly higher credit usage
Off
- · Instant response
- · Speaks from existing knowledge
- · No web search
- · Lower credit usage
Chat Folders
When your chat list gets long, it becomes hard to navigate. Folders let you group chats by topic. Create folders from the colored pill bar in the left menu, and move a chat into a folder from its three-dot menu.
- Create: The
+button on the folder bar in the left menu. Pick a name and color. - Move: Right-click a chat or use the three-dot menu and choose Move to folder. The folder list appears.
- Delete: Right-click the folder. When you delete a folder, the chats inside fall back into All — no chat is lost.
Knowledge Base
Upload your own PDF, Word, or text files so that the assistant answers from them. The assistant cites excerpts from your documents and notes page numbers — it does not fabricate.
Create a KB
From the left menu choose 'Knowledge Base' → 'New KB'. Enter a name and description.
Upload files
PDF, DOCX, DOC, TXT, MD — up to 50 MB per file. Multi-file upload is supported.
Wait until ready
The file becomes 'ready' within 1-3 minutes. Chunking and embedding happen in the background.
Attach to a chat
From the 'Choose KB' menu in the chat header, select your KB. It binds to that chat.
Ask a question
For every question, the assistant first searches your KB and answers citing the relevant excerpt.
Scheduled Tasks
Assign tasks that run automatically at specified times. Example: "Every morning at 9, summarize today's top tech news." The task runs on the server even when your computer is off; the result lands in your list as a new chat and optionally arrives as email.
Create a task
From the left menu choose 'Scheduled Tasks' → '+ New Task'.
Assistant + prompt + schedule
Pick an assistant, the prompt to be asked each time, and a schedule preset.
Notification preference
In-app notification and an email summary to your account address are enabled by default.
Wait (or 'Run Now')
The task triggers at the scheduled time. To test immediately, use 'Run Now'.
Result in chat + email
The chat is named 'Task: ...'; the email contains a summary + 'Continue in chat' button.
Deep Research
Some questions can't be answered in a single reply. Enable the purple "Deep Research" button in the chat header and ask; the assistant spends 5-10 minutes verifying multiple sources, presents conflicting evidence, and summarizes in a report format.
You can close the page; the research continues in the background. When done, you get an in-app notification and the report appears in your chat. A "Cancel" button is always on screen — you can stop and keep the partial result.
Canvas Editor
When you receive a long response — a blog draft, a code snippet, an email — click the "Open in Canvas" button below the message. The screen splits in two; the editor opens on the right. Select text or code and use the "Edit with AI" toolbar with ready commands (Shorter, More formal, Debug, Translate to Turkish...). Changes apply in place without disturbing the chat below.
- Auto-saves. You won't lose changes.
- Esc closes. Clicking outside the editor also closes it.
- Mode support. Markdown, code (JS/TS/Python/PHP/HTML etc.), plain text. Syntax highlighting included.
- Reflects to chat. Edits in Canvas are also saved back into the chat message; you see the up-to-date version after closing.
Regenerate Message
If you don't like the assistant's reply — want it shorter, more detailed, or a different angle — click the refresh icon below the message. The old answer is removed and a fresh one is generated for the same question. Your user message stays unchanged.
Advanced Memory
When the "Memory" toggle in the chat header is on, the assistant remembers things it learned about you in past conversations — your name, profession, preferred language, recurring projects. You don't start each chat from scratch.
Image Generation
Just tell the assistant "create an image of X" — it understands and generates one. When you upload an existing image, you can ask it to "change the color", "remove the background", "change the outfit" and similar commands (edit mode).