28a4b37dde
feat: showing the lists which an image is a part of
2025-07-29 15:54:51 +01:00
5024933852
feat: making the description markdown
2025-07-29 15:40:11 +01:00
a0bf27dd16
Haystack V2: Removing entities completely
2025-07-29 14:52:33 +01:00
3d05ff708e
feat: giving agent enough information to add to list instead of creating one
2025-07-29 12:14:45 +01:00
8597584cf0
feat: initial draft of generating a schema from one image
...
fix: error formatting
2025-07-29 11:37:23 +01:00
88d033314e
feat: initial attempt to create a schema
2025-07-29 09:47:59 +01:00
37f966e508
feat: finishing description agent infrastructure on backend
2025-07-24 14:12:41 +01:00
59bf884f5d
refactor: changing notes to be a simple image description
...
Notes would generate too often and not be very useful. This is much
better.
2025-07-24 13:59:24 +01:00
a283bc1bcd
feat: new AI generated lists
...
I think this could be how we generate other lists
Problems:
- Knowing it's a location is good because you can do nice stuff on the
frontend.
- Same for contacts & events.
So a good alternative, is to still use this type, but perhaps change the
database such that all lists live within the new tables (lists,
image_lists). But have special tags.
This would also make it easier on the AI I think.
2025-07-22 19:40:56 +01:00
e8a51ecc52
fix: the reason this wasn't working
...
This was also silent failing, so should probably have a look at it
2025-05-11 14:30:31 +01:00
5a2b990c0c
fix: adding extension hack
...
fix
2025-05-10 22:05:24 +01:00
6119938e52
mini just is better :(
2025-05-04 10:19:39 +01:00
b046a928b0
feat: using gpt-4.1-mini
...
feat: createExistingContact
feat: using nano instead of mini so I don't run out of money instantly
2025-05-03 18:25:35 +01:00
9860dd2dc5
feat(agents): providing a seed so it has more predictable results
2025-05-03 17:31:55 +01:00
a9ecd5818a
fix: re-requesting image when all agents are done
...
This means the agents are no longer parallel. Which will eventually be a
bottleneck, but I need to spend a bit more time on allowing for it to be
parallel, but now is not the time.
TODO: add a ticket
2025-04-27 14:45:23 +01:00
89ba950c5b
fix: enabling note agent
2025-04-19 10:30:49 +01:00
2b8e0695c6
feat(events): a better prompt with good integration with location agent
2025-04-19 10:30:29 +01:00
d448a41a9f
feat(contacts): not creating duplicates
2025-04-19 10:07:51 +01:00
a69d4e4d55
feat(location): prompt tweak + going back to faster model
2025-04-18 15:36:51 +01:00
6edc1e2915
feat(location): correctly updating an image if it contains a duplicate locatino
2025-04-18 15:32:07 +01:00
57f1e70c98
feat(agents): improving rationality by adding tool to allow the models to think through choices.
...
This works pretty nicely actually. I'm starting to understand how to
demistify the system prompt and have the tools the agent needs to do a
good job.
2025-04-18 15:06:20 +01:00
1b1f957e01
wip
2025-04-18 14:21:23 +01:00
49969b0608
feat(location-agent): using createLocation instead of updateLocation to simplify
2025-04-18 13:26:42 +01:00
9b95ffb59e
feat(contact-agent): using createContact with an ID field to provide updates
2025-04-17 18:57:13 +01:00
c9560f6881
feat(event-agent): update events function
2025-04-17 18:19:54 +01:00
c5535a5b3b
feat(location-agent): seperating the tool to allow for replying
...
This means it makes less mistakes and doesnt get as confused.
2025-04-17 18:09:00 +01:00
5ab0d13b21
fix(location-events): adding location id to the database from agent call
2025-04-17 15:32:50 +01:00
15289e4965
feat(prompts): adding better prompts & restoring tool_stop
...
Mistral's models seem to do something really strange if you allow for
`tool_choice` to be anything but `any`. They start putting the tool call
inside the `content` instead of an actual tool call. This means that I
need this `stop` mechanism using a tool call instead because I cannot
trust the model to do it by itself.
I quite like this model though, it's cheap, it's fast and it's open
source. And all the answers are pretty good!
2025-04-17 15:24:21 +01:00
181da1f09d
feat(orchestrator): removing the end tool call
...
fix
2025-04-17 13:00:39 +01:00
90b90a8185
chore: removing unnecessary logging
2025-04-17 13:00:24 +01:00
fb30eb4ad6
wip(orchestrator): improving orchestrator system prompt and tool description
2025-04-17 12:52:54 +01:00
5454a1cfaf
feat(event-location): communicating using tool calls correctly
2025-04-17 11:15:02 +01:00
3716d22eca
fix(logger): nil pointer error + log debug level clean
2025-04-17 11:07:37 +01:00
6d2f0c6108
refactor(agents): not returning an error on factory method
2025-04-17 11:02:11 +01:00
61c158d5b6
refactor(agents): encapsulating prompt and calls inside factory method
2025-04-17 10:58:19 +01:00
82331c0833
fix: using correct eventAgent instead of orchestrator bug + better logging
2025-04-17 10:48:30 +01:00
e42aa75639
refactor(agents): no need to wrap them in another struct
2025-04-17 10:36:11 +01:00
fa486153b4
feat: event agent calling location agent about location ID
...
This is pretty nice. We can now have agents spawn other agents and
actually get super cool functionality from it.
The pattern might be a little fragile.
2025-04-16 14:43:07 +01:00
aacecfffac
wip(agents): allowing event agent to call location agent
2025-04-15 16:44:00 +01:00
e89a342751
feat: Adding text message to describe an action3
2025-04-15 16:43:27 +01:00
e16b6f4529
fix
2025-04-14 20:08:07 +01:00
6ddae3426d
rollback: not using link functions as they are very problematic
2025-04-14 10:59:08 +01:00
8a57236f04
ffix
2025-04-14 10:40:02 +01:00
b138661991
prompt
2025-04-14 10:38:25 +01:00
51d36bf15b
more prompt
2025-04-14 10:36:21 +01:00
ecc2da5f86
fix more prompt
2025-04-14 10:33:56 +01:00
1f83b721a6
fix: prompts
2025-04-14 10:28:31 +01:00
5ae6a3403f
chore: removing old agent that was messy and too coupled
...
chore
2025-04-13 16:30:20 +01:00
3156cea904
feat(event): seperate event agent
2025-04-13 16:30:20 +01:00
d432d16752
feat(location): agent to create locations
2025-04-13 16:30:20 +01:00
