Museum Toolkit:
How we transformed artifact management for museums
Development in progress:
Seed Museums
N/A
Model generation time
-45%
Management time
-35%
Handle everything in dashboard
See their recent tasks they interacted with , visualize the stats of their quota usage, receive notification of their ongoing activities
Manage exhibition artifact
Generate 3D models
Build their own knowledge
Support from agent
New 3D Scanning
+
Editable knowledge-based AI agent
=
According to the 2004 Heritage Health Index (HHI), conducted by Heritage Preservation in partnership with IMLS, there are roughly 30,827 U.S. institutions responsible for preserving public trust collections.
Source
In the 2018 Museum Survey by Vernon Systems, 34% of respondents reported still using Excel to record collection data. It implies that approximately 10,500 organizations continue to rely on spreadsheets for at least part of their management.
Source
The same Vernon survey found that 77.59% of institutions used paper‐based systems, often in combination with other digital method to document their collections. It suggests that roughly 23,900 institutions still manage paperwork as a primary or supplemental record-keeping method.
Source
A large number of user groups
Rely on Inefficient digital management method
Highly rely on paper-based systems
The problems
A large number of US cultural relic management institutions, mainly museums, still heavily rely on Excel or paper-and-pencil recording methods for the management and archiving of cultural relics, as well as the planning of exhibitions, which leads to...
Archives and critical data are prone to loss
Low exhibition planning efficiency
Fragmented workflow hinder collaboration
How might we
Enable traditional museums and their staff to embrace digital and intelligent transformation beyond outdated workflows and tools?
Problem deconstruction
After settling HMW statement, we began to break down the problem.
Digital and Intelligent Transformation
=
Digital Transformation
+
Intelligent Transformation
=
Storing artifacts in a digital way
+
We can break down digital and intelligent transformation into digital transformation and intelligent transformation: A core function of transforming traditional museums into digital ones is storing artifacts in a digital way; while a core function of using digital management for museums is to leverage the power of AI to assist in museum management (including exhibition planning and general management).
Storing artifacts in a digital way
In the current information technology, there are 2D information storage and 3D information storage.
2D information storage
3D information storage
Compared to 2D information storage of cultural relics, 3D information storage can retain more details about the artifacts and also allows people to better visualize the relics themselves. Therefore, we believe that storing artifacts’ information in 3D form is a better choice.
Current restriction for 3D information storage
“It took our scanning specialist about 20 minutes to scan this object with Eva in HD, and then about 50 minutes to reconstruct it. Processing time in software took 110 minutes in total. The whole process took 2.5 hours.”
Source
“The fusion of technology and creativity can lead to remarkable results, but the learning curve can be steep; especially for beginners like me.”
Source
“Collecting a single point requires the CMM to move into position and that movement takes time. It’s common for a CMM inspection process to take 15–20 minutes – minimum – for it to move around and take the measurements.”
Source
Traditional 3D scanning is low-efficient
High learning curve
Take a lot of time for user to handle 3D scanning
How we solve it
Upload artifact videos, generate 3D models with built-in tools smoothly, and keep track of progress with task monitoring after that.
Organize their artifacts with exhibition-based 3D management and use tailored filters to streamline curatorial planning.
Utilize AI to help with museum management
Current pain point:
Small and medium-sized museum institutions hope to provide users with more comprehensive and high-quality services through the intervention of language models, but...
1
The mainstream large language models on the market are too generalized. Without fine-tuning or retrieval enhancement generation, they cannot effectively assist museums in providing services.
2
Building an agent customized based on the museum's situation is too cumbersome, and the museum also lacks the ability to sustain a team for this long-term.
How we solve it
We have decided to provide a quick two-step process that can be generically promoted, for the common tasks of museums, with functionalities built by small agents.
1
Quickly build a knowledge base for specific tasks
- To facilitate rapid setup, it supports multimodal input, and can also directly enter an information website address to complete the setup.
2
Build a functional agent based on the knowledge base
- In order to facilitate rapid construction, allow users to establish agents by utilizing our universal museum task model through multiple knowledge bases, without the need for additional operations.
Final evaluation
After building MVP, we conducted 50+ small-scale Business-end user testings, and the following are two quotes.
“The 3D scan appeared almost instantly and saved me at least 10 minutes compared to my previous photo-stitch workflow. Tagging was straightforward, though I missed the ‘condition’ field on my first try—an inline hint would help. The AI query returned exactly the right artifacts in under two seconds, which blew me away.”
“Generating the virtual tour script with the AI was incredibly fast and provided a clear, engaging narrative in under five seconds. Reordering tour stops proved awkward without a drag-and-drop timeline, which disrupted my creative flow. Exporting the final plan and voiceover text was seamless and saved me hours of manual scripting.”
This product was completed in design and delivered to the Development Team in Q2 of 2025, and the MVP development was also completed in Q2 of 2025. As a member of the Design Team, we plan to conduct a pre-launch user testing once the product is fully developed and collaborate with seed museum for deployment. Adjustments are expected to be made based on data feedback.